Article Structure
Abstract
We study the problem of generating an English sentence given an underlying probabilistic grammar, a world and a communicative goal.
Introduction
Suppose someone wants to tell their friend that they saw a dog chasing a cat.
Related Work
Two broad lines of approaches have been used to attack the general NLG problem.
Sentence Tree Realization with UCT
In this section, we describe our approach, called Sentence Tree Realization with UCT (STRUCT).
Empirical Evaluation
In this section, we compare STRUCT to a state-of-the-art NLG system, CRISP, 1 and evaluate three hypotheses: (i) STRUCT is comparable in speed and generation quality to CRISP as it generates increasingly large referring expressions, (ii) STRUCT is comparable in speed and generation quality to CRISP as the size of the grammar which they use increases, and (iii) STRUCT is capable of communicating complex propositions, including multiple concurrent goals, negated goals, and nested subclauses.
Conclusion
We have proposed STRUCT, a general-purpose natural language generation system which is comparable to current state-of-the-art generators.
Topics
natural language
- Natural language generation (NLG) develops techniques to extend similar capabilities to automated systems.Page 1, “Introduction”
- to Natural Language GenerationPage 1, “Introduction”
- This is then used by a surface realization module which encodes the enriched semantic representation into natural language .Page 2, “Related Work”
- In the MDP we use for NLG, we must define each element of the tuple in such a way that a plan in the MDP becomes a sentence in a natural language .Page 4, “Sentence Tree Realization with UCT”
- We have proposed STRUCT, a general-purpose natural language generation system which is comparable to current state-of-the-art generators.Page 9, “Conclusion”
See all papers in Proc. ACL 2014 that mention natural language.
See all papers in Proc. ACL that mention natural language.
Back to top.
generation process
- If so, we store it, and continue the generation process .Page 6, “Sentence Tree Realization with UCT”
- The times reported are from the start of the generation process , eliminating variations due to interpreter startup, input parsing, etc.Page 6, “Empirical Evaluation”
- Note that, as STRUCT is an anytime algorithm, valid sentences are available very early in the generation process , despite the size of the set of adjoining trees.Page 7, “Empirical Evaluation”
See all papers in Proc. ACL 2014 that mention generation process.
See all papers in Proc. ACL that mention generation process.
Back to top.
semantic representation
- Any combination which contains a semantic representation equivalent to the input at the conclusion of the algorithm is a valid output from a chart generation system.Page 2, “Related Work”
- This is then used by a surface realization module which encodes the enriched semantic representation into natural language.Page 2, “Related Work”
- For instance, a communicative goal of ‘red(d), dog(d)’ (in English, “say anything about a dog which is red.”) would match a sentence with the semantic representation ‘red(subj), dog(subj), cat(obj), chased(subj, obj)’, like “The red dog chased the cat”, for instance.Page 3, “Sentence Tree Realization with UCT”
See all papers in Proc. ACL 2014 that mention semantic representation.
See all papers in Proc. ACL that mention semantic representation.
Back to top.
